amansrivastava17/embedding-as-service

Supports multiple embedding models (BERT, XLNet, Word2Vec, etc.) with flexible pooling strategies (reduce_mean, reduce_max, etc.) to aggregate token embeddings into fixed-length sentence vectors. Deployable both as a Python module and as a client-server architecture with separate `embedding-as-service` server and `embedding-as-service-client` packages, enabling distributed inference across network boundaries. Built on transformer-based architectures with configurable sequence length and batch processing for production workloads.

210 stars. No commits in the last 6 months. Available on PyPI.